What Happens on Tuesday?

Abstract

This report presents data analytic outputs of Automated Traffic Signal Performance Measures (ATSPMs), a high-resolution signal and flow detector data from multiple intersections in Salt Lake City, released only for the Big Data Challenge on Signalized Intersections 2019. This study tried to analyze and present a new type of the high-resolution signal and flow detector data faster and easier as a bird’s-eye narrative for Messo-scopic level, aiming to support a decision-making process for current operators and stakeholders. It used R, an open-source language for data processing and visualization since R is one of the fastest and powerful tools for big-data processing and analytics. The report focused on presenting the patterns of performance measurements e.g., vehicle delay and signal time distributions for each intersection on a specific day of week and some of the results are presented via interactive maps and charts.

I. Introduction

To improve traffic performance measurements, the Utah Department of Transportation (UDOT) agreed to release a whole year 2018 ATSPMs dataset, consisting of 22 signalized intersections along two major corridors in Salt Lake City, Utah, the US. Each intersection along with two major corridors, 300 West and 700 East (see Figure 1 below) has its unique identification number. The high-resolution dataset with 0.1-second time interval is structured in a simple numeric text data but associated look-up tables and data dictionary require a careful interpretation not only to fathom a specific meaning of each numeric value e.g., vehicle, pedestrian detections, signal phase changes including rings and barriers, etc. but also to see the whole picture of the dataset. The size of ATSPMs data are relatively smaller than other traditional detector types, such as loop-detectors or blue-tooth-detectors thank to an effective utilization of the series of lookup tables (Sturdevant et al., 2012). Regardless of the data size, the ATSPMs data allows to explicate unconventional stories about traffic and signal patterns in Meso-scopic level.

Figure 1. 300 West and 700 East of Salt Lake City

II. Literature Review

A few of performance measurements studies investigated the hi-resolution traffic signal data. Day et al. (2011) developed a series of optimization functions using site-observed high-resolution detector data to adjust offsets about an arterial system of eight coordinated signals in Noblesville, Indiana. The study utilized green time, arrivals per cycle, and estimated queue size per cycle to minimize delays and stops. Brennan et al. (2011) created a series of visualization frames to provide educational insights on signal systems under different parameter settings and system behaviors; the list of visualization frames includes time-of-day schedule change time, observed cycle length, green time/split time, coordinated phase actuation, early return to green, arrivals over advance detection relative to green indication, progression quality characteristics related to offset, adjacent signal synchronization, coordinated phase operation in rest, plan time changes, preemption, impact of queuing, and longitudinal analysis of splits. Recently, Federal Highway Administration (FHWA) began to promote the utilization of ATSPMs, indicating the importance of understanding and interpreting these high-resolution data. To date, 11 state department of transportation (DOTs) and 26 transportation agencies in the US. are already pursuing this next generation of traffic signal operating system (FHWA, access date May 21, 2019). Some of the agencies conducted case studies and measurements using ATSPMs (Kimble 2017 and Gault 2018).

III. Data Structure and Processing

All observations from each detector are coded with four categories: SignalID, Time, Event Code, and Parameter. SignalID indicates a geographical location for each detector and it becomes a key piece of information for detector decoding, such as approach, detectors, movement and lane information. Time is a recording time for each observation. Event Code with SignalID tells whether a detection is derived from vehicle movement or signal changes. Lastly, Parameter explains a detail of each Event Code such as phase information for signal detection and a feature for vehicle detection. Below figure describes the overall data architecture of the high-resolution detector data.

Figure 2. A general view for data structure

Data pre-processing was performed by splitting signal and detector information for each record as well as linking look-up tables and data dictionary for appropriate detail information.

a. Signal

Signal timing information from Table 1, out of 22 look-up tables, was used for estimating signal allocation distribution. Based on NEMA phase numbers at UDOT, signal time distribution was calculated by phase and month. This study focused to see how long signal time allocation was distributed to each phase at each intersection.

Table 1. Key signal information, used for this study

b. Detector

The detector data contains vehicle movement information by time, lane and occupancy. However, the dataset found to be incomplete. Some of Event Code information was missing. A corresponding value of the Event Code was often missed in the lookup table(s) as well. Therefore, this study focuses on a snapshot of operational performance in Messo-scopic level analysis instead of a Micro-scopic level. Given limitation of data and resources, this study picked specific day of week, i.e., Tuesday as it showed the highest traffic observation in weekdays. About 17% of weekly traffic was observed on Tuesday. See Figure 3 showing the average vehicle observations by day of week along two corridors, estimated by any observations for detector-off (Event Code #81) after removing duplicated records.
(actual traffic counts can be seen by hovering mouse over the bar chart)

Figure 3. Average observations by day of week

This study focuses on 3000-West corridor. A same analysis and approach can easily apply to the other corridor(s) or day(s).

IV. Methodologies

The study applied relatively simple methodologies. This section briefly describes them for both signal and detector data.

a. Signal allocation distribution

Signal timing information from Table 1 was used for obtaining signal allocation distribution. Based on NEMA phase numbers at UDOT, signal time distribution was calculated by phase and month. This study focused to see how long signal allocation was distributed to each phase at each intersection.

b. Vehicle delay distribution

For vehicle detector processing at each intersection, vehicle delay was estimated by 1) a vehicle dwell time between every first ‘detector-on’ and last ‘detector-off’; i.e., Event Code #81 and #82. A vehicle count for each approach and lane was estimated by 2) a pair of count between every last ‘detector-on’ and first ‘detector-off’. As aforementioned, some of Event Codes in detector data was missing in the detector look-up table, which resulted in removal of records for a final analysis. Finally, the processed vehicle movement data was obtained by lane, time of day and direction.

V. What happens in Tuesday?

a. Signal allocation distribution

A series of figures below show signal phase distributions at each intersection on Tuesday from January 2018 to December 2018. These phase numbers are referred by NEMA Phase Number Convention at UDOT. A distribution ratio for each phase was obtained by examining monthly signal allocation frequencies for each phase. According to the series of these snapshots, 700-East corridor overall shows consistent signal distributions for a major direction, i.e., North bound and South bound (except for Signal ID # 7076), whereas 300-West corridor shows relatively mixed signal distributions. For example, some of intersections along 300-West corridor between Signal ID # 7122 and Signal ID # 7124 disclosed mostly major directional-bound signal distribution. However, Signal ID # 7125 to southbound (till Signal ID # 7241) showed diverse signal distributions due in part to other heavy approaching corridors, such as North Temple, 400-South corridor.

(note: Zoom-in/zoom-out with the mouse and click to see detailed numbers for each intersection)

Jan

Figure 4. Signal distribution snapshot by phases by month

Feb

Figure 4. Signal distribution snapshot by phases by month

Mar

Figure 4. Signal distribution snapshot by phases by month

Apr

Figure 4. Signal distribution snapshot by phases by month

May

Figure 4. Signal distribution snapshot by phases by month

Jun

Figure 4. Signal distribution snapshot by phases by month

Jul

Figure 4. Signal distribution snapshot by phases by month

Aug

Figure 4. Signal distribution snapshot by phases by month

Sep

Figure 4. Signal distribution snapshot by phases by month

Oct

Figure 4. Signal distribution snapshot by phases by month

Nov

Figure 4. Signal distribution snapshot by phases by month

Dec

Figure 4. Signal distribution snapshot by phases by month

b. Vehicle delays

Based on the detector data, some of informative delay records were observed by time of day, month of year, and others. Some of the findings are presented here. Figure 5a and 5b show observed average delays for each intersection in different time of days. The first figure, 5a, shows the Northbound observations, while 5b shows the Southbound observations. Except for extremely high observations in #7123 in Southbound, three intersections: #7127, #7241 and #7126 are the highest delay intersections observed in both bounds. Overall, afternoon peak, between 15th-19th hour (in military hour), revealed more average delays than morning peak, i.e., 6th- 8th hour. Also, Northbound disclosed relatively higher average delays compared to Southbound in morning peak, and afternoon peak.

cat(“”)

c. Vehicle Movements and Delay

Using the traffic counts and delays at each intersection by movement and approach, a simple but interactive traffic count map is presented in following Figure 6. Each circle bar in the map represents an average of total delay by multiplying each count and average delay at each intersection. Intersection #7122 shows a quite mixed traffic pattern across the movements in January 2018, e.g., right, through and right. Traffic estimates in intersection #7123, #7127 and #7129 were low due to lack of detector data records and/or missing look-up table information. Two intersections: #7125 and #7241 showed the highest average total delay (total delay = counts x average delay) amongst all intersections.

(note: you can hover to check actual average total delay for each intersection and click to see actual observed traffic counts for each movement)

Jan

Figure 6. Directional Traffic Counts and Delays in Jan

Feb

Figure 6. Directional Traffic Counts and Delays in Feb

Mar

Figure 6. Directional Traffic Counts and Delays in Mar

Apr

Figure 6. Directional Traffic Counts and Delays in Apr

May

Figure 6. Directional Traffic Counts and Delays in May

Jun

Figure 6. Directional Traffic Counts and Delays in Jun

Jul

Figure 6. Directional Traffic Counts and Delays in Jul

Aug

Figure 6. Directional Traffic Counts and Delays in Aug

Sep

Figure 6. Directional Traffic Counts and Delays in Sep

Oct

Figure 6. Directional Traffic Counts and Delays in Oct

Nov

Figure 6. Directional Traffic Counts and Delays in Nov

Dec

Figure 6. Directional Traffic Counts and Delays in Dec

VI. Conclusion and Comment

This study presented Messo-scopic snapshots of performance measurements using open source, R. Unlike traditional space-time diagrams or effective green time cumulative curves (Day et al. 2011; Steve, 2017; Steve, 2018), it focused on the bird-eye view of a corridor-level performance measurements. This study presented several performance measurements such as signal allocation distribution, vehicle delay distribution and vehicle movement and delay at each intersection and movement type. This study did not include pedestrian related performance measurements due to restricted time and limited resources. However, they can be easily performed by applying a same logic applied in this study. This study still has a key drawback, including missing micro-level performance measurements analysis such as stopped delay, movement delay by lane and movement at each intersection. They are the study topics that I will explore in the future.

The emerging technology of detector and detailed automated traffic signal performance measures (ATSPMs) brings enlarged operational benefits. For the full benefit, further attention into the data, especially data integrity, needs to be paid.

Reference

[1] Day, C.M., Brennan Jr, T.M., Hainen, A.M., Remias, S.M., Premachandra, H., Sturdevant, J.R., Richards, G., Wasson, J.S. and Bullock, D.M., 2011. Reliability, Flexibility, and Environmental Impact of Alternative Arterial Offset Optimization Objective Functions. Transportation Research Record.

[2] Brennan Jr, T.M., Day, C.M., Sturdevant, J.R. and Bullock, D.M., 2011. Visual Education Tools to Illustrate Coordinated System Operation. Transportation Research Record, 2259(1), pp.59-72.

[3] Sturdevant, J.R., Overman, T., Raamot, E., Deer, R., Miller, D., Bullock, D.M., Day, C.M., Brennan Jr, T.M., Li, H., Hainen, A. and Remias, S.M., 2012. Indiana traffic signal hi resolution data logger enumerations.

[4] Steve Kimble, 2017, Leveraging Hi-Res Data for Signalized Corridor Monitoring, ITS Midwest Conference.

[5] FHWA, Automated Traffic Signal Performance Measures (ATSPMs), (https://www.fhwa.dot.gov/innovation/everydaycounts/edc_4/atspm.cfm: access date 21/May/2019)

[6] Steve Gault, 2018, Automated Traffic Signal Performance Measures, Penn State Engineering and Safety Conference.